Distributed Frameworks for Dynamic Human/Agent Collaborative Problem Solving

Period of Performance: 06/19/2008 - 09/24/2010

$750K

Phase 2 SBIR

Recipient Firm

Cougaar Software, Inc.
7600 Leesburg PrikeEast Building - Suite 105
Falls Church, VA 22043
Principal Investigator

Abstract

In the Phase I effort, we studied the problem of how to achieve a high-quality distributed system for dynamic human / agent collaborative problem solving in a specialized domain. From that effort we determined that context was a key missing element that made that collaboration difficult to establish and maintain. Based on our work in Deep Situational Reasoning, we derived a framework from which an agent can maintain context with the human as an integral part of the user interface. Specifically, it combined the classical Model-View-Controller interface paradigm with our newer Object-Network-Semantic paradigm. The combination provided an interesting fusion of meaning, relationship, situated state, environment and intentions. It is our goal in this phase to build a working prototype of this framework and demonstrate it in a limited problem domain